A unified web extraction and stateful browser automation engine for AI agents, replacing heavy testing frameworks.
cortex-scout is worth checking the docs before setup with strong trust signals. Check agent compatibility and use-case fit before adding it to your workflow.
gh repo view cortex-works/cortex-scout --webOpen the official repository or website.
Check the README for package manager, auth, and platform requirements.
Try it in a small test task inside your agent workflow.
CortexScout is a tool that helps AI agents search the web, extract content, and control a browser automatically. It handles tricky things like bot detection and JavaScript-heavy pages, and can even ask a human for help when needed. It's designed to be efficient and work well with large language models.
CortexScout is a self-hostable Rust binary that provides search, extraction, and stateful browser automation capabilities over MCP (stdio) and an optional HTTP server. It is built to handle practical failure modes of web retrieval such as rate limits, bot challenges, and JavaScript-heavy pages through progressive fallbacks: native retrieval → Chromium CDP rendering → stateful E2E testing → Human-in-the-Loop (HITL) workflows. It includes a built-in stateful CDP automation engine designed specifically for AI agents, replacing heavy frameworks like Playwright or Cypress for E2E testing. Key features include a unified tool family for search, fetch, crawl, and extraction; anti-bot handling with proxy rotation and block-aware retries; HITL support for CAPTCHA or complex logins; memory-backed research history via LanceDB; and deep research capabilities for multi-hop search and synthesis. It integrates with the Cortex-Works ecosystem for multi-agent scaling and shared retrieval artifacts.
Strong trust signals; still review the README and permissions before production use.
Last commit was about 57 days ago.
66 GitHub stars indicate community interest.
1 open issues signal maintenance load.
MIT license detected.
AI agents performing deep research on the web with multi-hop search and synthesis.
Automated E2E testing of web applications without heavy frameworks like Playwright.
Extracting structured data from JavaScript-heavy or bot-protected websites.
Building knowledge graphs by crawling and extracting content from multiple pages.
Human-in-the-loop workflows for handling CAPTCHAs or complex authentication during web automation.
Automated browser control may violate website terms of service; use responsibly.
Storing cookies and session data persistently could pose privacy risks if not secured.
Proxy rotation and anti-bot bypass techniques may be considered unethical if used for scraping without permission.
66
Stars
7
Forks
1
Issues
MIT
License
Browser automation CLI built for AI agents — breaks anti-bot walls, supports multi-session parallel execution.
Anti-detection browser server for AI agents with C++ engine-level fingerprint spoofing.
MCP server for browser automation using Playwright, enabling LLMs to interact with web pages.
3 security/trust notes recorded.
Setup difficulty is 3/5.